Specialized interferon ligand action in COVID19

The impacts of IFN signaling on COVID19 pathology are multiple, with protective and harmful effects being documented. We report here a multi-omics investigation of IFN signaling in hospitalized COVID19 patients, defining the biosignatures associated with varying levels of 12 different IFN ligands. Previously we showed that seroconversion associates with decreased production of select IFN ligands (Galbraith et al, 2021). We show now that the antiviral transcriptional response in circulating immune cells is strongly associated with a specific subset of ligands, most prominently IFNA2 and IFNG. In contrast, proteomics signatures indicative of endothelial damage associate with levels of IFNB and IFNA6. Differential IFN ligand production is linked to distinct constellations of circulating immune cells. Lastly, IFN ligands associate differentially with activation of the kynurenine pathway, dysregulated fatty acid metabolism, and altered central carbon metabolism. Altogether, these results reveal specialized IFN ligand action in COVID19, with potential diagnostic and therapeutic implications.


INTRODUCTION
The impact of Interferon (IFN) signaling on the course of COVID19 pathology has been the subject of much investigation, with both protective and deleterious effects being reported. The protective effects of IFN signaling are demonstrated by studies showing that severe COVID19 is associated with decreased IFN signaling (1), the presence of auto-antibodies blocking IFN ligand action (2), and genetic variants that impair IFN signaling (3). However, Type I IFN signaling has been established as a driver of pathology in mouse models of both SARS-CoV-1 and SARS-CoV-2 infections (4, 5), and Type I and III IFN signaling have been implicated in disruption of lung barrier function and increased susceptibility to secondary bacterial infections in mice (6). This ambivalence has fueled the design of seemingly contradictory clinical trials using either IFN ligands (7) or agents that block IFN signaling, such as JAK inhibitors (8). This duality is further illustrated by studies showing that genetic variants leading to low expression of the Type I IFN receptor IFNAR2 or high expression of TYK2, a protein kinase required for Type I IFN signaling, are associated with life-threatening disease (9). Therefore, it is possible that context-dependent variations in IFN signaling may attenuate or exacerbate COVID19 pathology in different settings. Indeed, retrospective analysis of IFN-α2b treatment in COVID19 showed that early administration was associated with reduced mortality, whereas late administration was associated with increased mortality (10).
There are three major types of IFN signaling defined by the transmembrane receptors and downstream signaling kinases involved (11). Type I IFN involves IFN alpha, beta, epsilon, kappa, and omega ligands, the IFNAR1 and IFNAR2 receptors, and the downstream kinases JAK1 and TYK2. Type II IFN signaling involves the gamma ligand, the IFNGR1 and IFNGR2 receptors, and the downstream kinases JAK1 and JAK2. Type III IFN signaling involves the lambda ligands, the IFNLR1 and IL10RB receptors, and the JAK1 and TYK2 kinases. However, this classification fails to capture the biological complexity created by the differential effects of distinct ligands acting through the same receptors. This is most evident by the differential effects of alpha ligands and IFNB1 within Type I signaling (12). Even within alpha ligands there is significant heterogeneity in cellular source, site of action, and downstream effects (12). Nevertheless, in the context of lung viral infections, it is accepted that select alpha and lambda ligands are first responders in the antiviral response due to their induced expression upon engagement of pattern recognition receptors in the lung epithelium (13). In the context of SARS-CoV-2 infections, little is known about the functional specialization of different IFN ligands and their relative contributions to different aspects of the ensuing pathology. Furthermore, SARS-CoV-2 has evolved diverse strategies to evade IFN signaling (14), and clinical trials for IFN alpha, beta, gamma and lambda ligands have been completed or are under way, even in combinations, but definitive results leading to approval for clinical use are pending (15).
Within this context, we report here a multi-omics analysis of IFN signaling in hospitalized COVID19 patients. This investigation includes a comprehensive examination of the whole blood transcriptome, plasma proteome, anti-SARS-CoV-2 antibodies, peripheral immune cell repertoire, and plasma and red blood cell metabolomes in relationship to levels of 12 different circulating IFN ligands. In hospitalized patients with moderate COVID19 pathology, transcriptome-based IFN scores are highly variable and significantly associated with levels of a subset of circulating IFN ligands such as IFNA2 and IFNG, but not so IFNA6 or IFNB1. Likewise, plasma proteomic signatures are also differential among ligands. For example, whereas IFNG and other ligands are clearly associated with production of monocyte activating and mobilizing chemokines, IFNA6 and IFNB1 levels associate with markers of platelet degranulation and endothelial damage. Furthermore, IFN ligands display differential relationships with immunoglobulins targeting SARS-CoV-2, revealing that seroconversion associates with decreased production of a select subset of ligands. This shift in IFN ligand production upon seroconversion is accompanied by significant changes in the immune cell types associated with production of the various ligands. For example, whereas IFNA10 is strongly associated with levels of Th1 CD4 T cells, CD56 bright NK cells and plasmacytoid dendritic cells, its levels are strongly anti-correlated with levels of circulating plasmablasts. Lastly, we revealed specific metabolomic signatures associated with diverse ligands.
Whereas IFNG is the most strongly associated with tryptophan catabolism through the kynurenine pathway, other ligands associate with metabolic pathways indicative of dysregulated central carbon metabolism, nitric oxide metabolism, and fatty acid oxidation. Altogether, these results indicate that purified recombinant proteins were commercially available, into a pooled plasma reference sample (only IFNW1 could not be obtained, see Methods). This test led us to discard five SOMAscan ® measurements (IFNA5, IFNA8, IFNA14, IFNA21, IFNL2) due to lack of sensitivity, and to relabel some measurements based on apparent cross-reactivity, such as IFNA4/16, IFN7/17/21, and IFNL3/2 ( Figure 1 -supplement 1B). When the same IFN ligand was measured by both platforms, we preferred the MSD measurements, which are quantified against a standard curve (Figure 1supplement 1C). This exercise allowed us to focus on measurements for 12 IFN ligands in our subsequent analyses: IFNA1, IFNA2, IFNA4/16, IFNA6, IFNA7/17/21, IFNA10, IFNA16, IFNB1, IFNG, IFNL1, INFL3/2, and IFNW1 (Figure 1 -supplement 1B-D). We next determined Spearman correlations between the RNA-based IFN Alpha transcriptional scores and levels of the 12 IFN ligands ( Figure 1D). Interestingly, the correlations were highly variable, and four of the ligands did not show significant associations with the IFN Alpha transcriptional scores (IFNB1, IFNA16, IFNW1, and IFNA6).
This result is clearly illustrated by the Type I ligands IFNA2 and IFNA6, which are the most and least correlated with IFN Alpha scores, respectively. Although both ligands are significantly upregulated in the plasma of COVID19 patients (Figure 1E), only IFNA2 levels correlate with the IFN Alpha scores ( Figure 1F) and with expression of well recognized ISGs, such as ISG15 and OAS2 (Figure 1F-G).
These differences could not be simply explained by the degree of induction of the various ligands in COVID19 patients, as illustrated by IFNA7/17/21, IFNL3/2, IFNA10 and IFNL1, all of which were not statistically higher among COVID19 patients (Figure 1 -supplement 1D) but nonetheless correlated significantly with IFN Alpha scores ( Figure 1D). Repeating this analysis for IFN Gamma scores produced a very similar rank of correlations, which is perhaps not surprising given the high overlap between the IFN Alpha and Gamma Response Hallmark GSEA gene sets (18) (Figure 1 -

supplement 1A, see Methods).
To explore this phenomenon more deeply, we completed a comprehensive analysis of gene expression signatures in the whole blood transcriptome associated with varying plasma levels of the 12 IFN ligands, using only data from COVID19 patients. Toward this end, we defined Spearman correlations between each ligand and mRNAs for 15,000+ genes detected by RNAseq, which identified thousands . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint of significant positive and negative correlations, with great variability across ligands (Figure 1supplement 2A, Supplementary file 4). We then analyzed the ranked correlations for each IFN ligand using GSEA to identify known gene sets with significant enrichment among positive or negative correlations with the levels of each ligand ( Figure 1H, Supplementary file 5). This analysis showed that the top gene signatures positively associated with 8 of the IFN ligands are indeed the IFN Alpha and Gamma Responses, followed by related inflammatory and immune pathways (e.g. TNFA signaling, Inflammatory response, IL6/JAK/STAT3 signaling). In contrast, for the other 4 ligands (IFNB1, IFNA6, IFNW1, and IFNA16), the top signatures enriched in the positive correlations are related to cell proliferation, such as G2M checkpoint, E2F targets, and MYC targets ( Figure 1H). In fact, some of these ligands show negative correlations with the IFN Alpha and Gamma Responses ( Figure 1H). This result is once again illustrated by the differential behavior of IFNA2 and IFNA6. Whereas mRNAs positively associated with IFNA2 show clear enrichment of the IFN Alpha Response gene set, these same mRNAs are negatively correlated with IFNA6 levels (e.g. ISG15, Figure 1 -supplement 2B).
Altogether, these results suggest functional specialization among circulating IFN ligands, whereby only a fraction of ligands associates with the recognizable IFN transcriptional response in circulating immune cells.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint Figure 1. IFN signaling at the whole-blood transcriptome level correlates with a subset of IFN ligands. (A) Volcano plot for DESeq2 differential expression analysis of gene-level count data for COVID19-positive vs. -negative samples, adjusted for age and sex. Horizontal dashed line indicates an FDR threshold of 10% for negative binomial Wald test; numbers above plot indicate significant genes at this threshold. Interferon stimulated genes (ISGs) are highlighted in green. (B) Bar plot of top 10 Hallmark gene sets as ranked by absolute normalized enrichment score (NES) from Gene Set Enrichment Analysis (GSEA). Bar color represents NES; Bar length represents -log10(FDR q-value). (C) RNA-based IFN Alpha scores, separated by COVID19 status. Scores were calculated for each research participant by summing Z-scores for 51 differentially expressed genes from the Interferon Alpha Response Hallmark gene set from MSigDB. Z-scores were calculated from the adjusted concentration values for each gene in each sample, based on the mean and standard deviation of COVID19-negative samples. Data are presented as a modified sina plot with box indicating median and interquartile range. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint IFN ligands show differential proteomic signatures associated to COVID19 pathophysiology.
Next, we investigated the proteomic signatures associated with each ligand. Using a linear model adjusting for age and sex, we identified 963 epitopes measured by the SOMAscan ® platform differentially abundant in the plasma of COVID19 patients (Figure 2A, Supplementary file 6). GSEA identified Hallmark IFN Alpha and Gamma Responses as the top proteomic signatures induced in COVID19 ( Figure 2B, Supplementary file 7). As for the transcriptome, we created protein-based IFN alpha and gamma scores for each participant, which showed significantly higher yet highly variables IFN scores among COVID19 patients ( Figure 2C, Figure 2 -supplement 1A). Notably, plasma protein-based IFN scores may inform about the organismal IFN response, not just that of circulating immune cells driving the whole blood transcriptome IFN signature, as multiple organs and tissues could contribute to secretion of IFN-related proteins. We then defined correlations between the 12 ligands and the protein-based IFN scores, which revealed some similarities and differences relative to the RNA-based IFN scores ( Figure 2D). Whereas IFNA2 and IFNG remained as the ligands most  Figure 2E). This suggest that whereas IFNA6 and IFNB1 may not contribute to the IFN transcriptional response of circulating immune cells, they may . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint nonetheless contribute to IFN responses in peripheral tissues and organs contributing to the proteinbased plasma IFN signature. This is illustrated by the behavior of CXCL11 (IFN-inducible protein 9), a canonical ISG, which is significantly correlated at the protein level with IFNA2, IFNA6 and IFNB1 (Figure 2 -supplement 2A). Additionally, IFN ligands often display highly dissimilar, even opposite, relationships to certain proteomics signatures. This is clearly illustrated by the PI3K/AKT/mTOR signature, which was positively correlated with some ligands and negatively correlated with others ( Figure 2E, compare correlations to HRAS for IFNA1, IFNA6 and IFNB1 in Figure 2 -supplement 2B).
To probe further into this phenomenon, we examined the top 5 positively and negatively correlated epitopes for each ligand using unsupervised clustering analysis, which revealed many specialized relationships with potential relevance to COVID19 pathophysiology (Figure 2 -supplement 2C). For example, several chemokines involved in immune control showed differential associations, such as CXCL10 (IP10, compare IFNG to IFNW1 in Figure 2F); CX3CL1 (fractalkine, compare IFNA10 to IFNA6 in Figure 2G); CCL7 (MCP3, compare IFNA2 to IFNA16 in  Figure 2H). This suggests that IFNB1 production is associated with platelet activation, which could be interpreted as a sign of endothelial damage at sites producing IFNB1. A subset of IFN ligands showed strong associations with components of the complement cascade, such as C1QC (Figure 2 -supplement 2C, compare IFNA2 to IFNB1 in Figure 2I). The top correlated epitope for IFNA10 is TRIL is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint ligands, most prominently IFNA6, but not others (Figure 2 -supplement 2C, compare IFNA6 to IFNA10 in Figure 2 -supplement 2G). OLFM4 (Olfactomedin 4), a protein selectively expressed in inflamed colonic epithelium (19), was strongly associated with IFNA4/16, but not other ligands ( Figure   2 -supplement 2C, compare IFNA4/16 versus IFNA10 in Figure 2 -supplement 2H).
Altogether, these results reveal that circulating levels of different IFN ligands associate with proteomics signatures indicative of multiple pathophysiological processes, such as tissue-specific inflammation, complement activation, and endothelial damage.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint model fit with 95% confidence intervals in grey. (C) Heatmap representing correlations between plasma levels of proteins measured by SOMAscan ® and each IFN ligand. Values displayed are Spearman correlation scores (Rho) for proteins ranked in top 5 positive or top 5 negative correlations for at least one IFN; asterisks indicate significant correlations (10% FDR); columns and rows are grouped by hierarchical clustering.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Differential relationship between IFN ligands and seroconversion status.
Next, we analyzed correlations between the 12 IFN ligands and the MS plasma proteomics dataset.
The MS proteomics platform is highly complementary to the SOMAscan ® dataset, as it detects many abundant proteins for which SOMAmer ® reagents are not available, such as various immunoglobulins (Igs). Using a linear model adjusting for age and sex, we identified 70 proteins differentially abundant in the plasma of COVID19 patients (Figure 3 - Figure 3D). Lastly, IFN ligands have clearly distinct relationships to a subset of immunoglobulin heavy and light chain variable domain peptides, that were either strongly positively or negatively regulated with the levels of . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; specific ligands (compare IFNA2 to IFNA6 in Figure 3E and IFNA1 to IFNB1 in Figure 3F). This result could potentially be explained by varying levels of IFN ligands upon seroconversion (16).
In order to investigate in detail the interplay between specific IFN ligands, immunoglobulin expression, and seroconversion, we examined correlations between the ligands and all immunoglobulin variable domains detected by MS proteomics, as well as seroconversion assays used to detect IgGs against SARS-CoV-2 peptides (S1 full length, spike; S1 N-terminus; and S1 receptor binding domain, RBD; nucleocapsid) ( Figure 3G). This analysis revealed that a subset of IFN ligands is strongly anticorrelated with seroconversion (e.g., compare IFNA2 to IFNB1 in Figure 3H) and specific Ig variable domains that have been previously found enriched in the bloodstream of COVID19 patients, such as IGHV1-24 and IGLV3-1 (20, 21). This could be interpreted as early production of some ligands which subsequently declines with seroconversion (e.g., IFNA2, IFNG), followed by later production of other ligands (e.g IFNA6, IFNB1), potentially from sites where SARS-CoV-2 evades humoral neutralization.
Overall, these results further support the notion of differential action of IFN ligands in COVID19 pathophysiology, suggesting a temporal sequence of IFN production prior and after seroconversion.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Figure 4B, compare IFNA10 to IFNA16 in Figure 4C). This pattern was also apparent for many, but certainly not all, T cell subsets ( Figure 4B). Similarly, NK CD56 bright cells also showed differential positive relationships with IFN ligands, with an overall pattern similar to that of key T cell subsets (compare IFNA2 to IFNA16 in Figure 4D). Notably, this analysis also revealed significant positive associations between specific ligands and plasmacytoid dendritic cells (pDCs), . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
Many of the IFN ligands positively associated with CD4+ T cell subsets were negatively associated with B cell subsets, while IFNA6 displays the opposite relationship ( Figure 4B). This is clearly illustrated by plasmablasts (compare IFNA10 versus IFNA6 in Figure 4F). These differential associations could be interpreted as a transition from innate T cell-driven responses prior to seroconversion, followed by B cell activation and differentiation toward antibody-producing plasmablasts during seroconversion, along with decreased production of a specific subset of IFN ligands.
Altogether, these results suggest a temporal sequence of IFN ligand production in coordination with changes in the peripheral immune cell compartment, whereby a larger subset of ligands is produced early on during the innate immune response, whereas a few others are associated with development of the adaptive humoral response. An overview of salient IFN ligand associations along the paths of T cell and B cell activation and differentiation is shown in Figure 4 -supplement 2B.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Metabolic signatures of IFN ligand action.
Next, we investigated metabolic signatures associated with varying levels of IFN ligands, calculating Spearman correlations for detected metabolites in plasma and red blood cell (RBC) samples against each of the IFN ligands (Figure 5A-B, Figure 5 -supplement 1A-B, Supplementary files 14-15, see Methods). In plasma, significant positive correlations were observed between the tryptophan/indole pathway metabolites kynurenine and 5-hydroxyindoleacetate and IFNG, but not other IFNs (compare IFNG to IFNA16 in Figure 5C). In RBCs, kynurenine levels (as well as those of indoxyl, another tryptophan/indole metabolite) showed a strong positive association with IFNG, as well as IFNA7/17/21 (compare IFNG to IFNA7/17/21 in Figure 5D). Activation of the kynurenine pathway has been well documented in COVID19 (25)(26)(27)(28)(29). Kynurenine production can be stimulated by induction of IDO1 (indoleamine-2,3-dioxygenase 1), an ISG downstream of all three major types of IFN signaling (30).
Therefore, it is interesting that this pathway is preferentially associated with IFNG in COVID19.
Plasma levels of IFNA2 showed significant positive correlations with the markers of oxidative stress glutathione disulfide and 5-oxoproline, a byproduct of the gamma-glutamyl cycle (Figure 5A, compare IFNA2 to IFNW1 in Figure 5 -supplement 1C), and negatively associated with markers of endothelial dysfunction and nitric oxide signaling (arginine, citrulline) ( Figure 5A, compare IFNA2 to IFNW1 in Figure 5 -supplement 1D), as well as metabolites of potential bacterial or iatrogenic origin (mannitol) and derived from purine oxidation (hypoxanthine) (Figure 5A). In RBCs, IFNA2 had once again strong positive correlations with several markers of oxidative stress (5-oxoproline) or pentose phosphate pathway activation (sedoheptulose phosphate) (Figure 5B, compare IFNA2 to IFNB1 in Figure 5E), which is required in RBCs to generate reducing equivalent (NADPH) for recycling oxidized glutathione and other NADPH-dependent antioxidant enzymes. IFNA2 levels also positively correlated with fatty acid mobilization in RBCs-perhaps as a result of the activity of peroxiredoxin 6 (31) or phospholipase A2 activity (32, 33) on complex lipids to fuel fatty acid release in the bloodstream to sustain viral capsid formation (34). Of note, among the positive correlates to IFNA2 levels in the fatty acid compartment, we observed only saturated (octanoic, dodecanoic, hexadecenoic, octadecanoic) or monounsaturated fatty acids (tetradecenoic, hexadecenoic, octadecenoic) (compare IFNA2 to IFNB1 in Figure 5F), suggestive . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint of limited fatty acid desaturase activation despite the stress induced by the viral infection (35). Several ATP precursors/breakdown products (AMP and adenine) positively correlated with IFNA2 in RBCs, as did pyruvate, phosphate and diphosphate -suggestive of altered glycolysis and overall energetics as a function of IFNA2 signaling. IFNA2 also negatively correlated with several amino acids in RBCs, including the antioxidants taurine, arginine, threonine and methionine -critical for RBC redox damage repair in the face of the incapacity to synthesize new proteins (36).
Plasma IFNL1 significantly correlated with several glycolytic metabolites (e.g. pyruvate, compare IFNL1 to IFNA1 in Figure 5G . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint Altogether, these results not only confirm metabolic signatures previously associated with IFN signaling (e.g. activation of the kynurenine pathway), but also reveal unexpected associations between specific IFN ligands and diverse metabolic processes with ties to COVID19 pathophysiology.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021.  (43)(44)(45)(46)(47)(48)(49). In the context of COVID19, the role of IFN signaling has been the subject of much study and debate, with both protective and deleterious effects being documented in different experimental systems and clinical settings (2-4, 10, 14, 15). Within this framework, we provide here a comprehensive analysis of multi-omics signatures associated with production of multiple IFN ligands in hospitalized COVID19 patients, revealing a high degree of diversity and specialization, even among ligands in the same subfamily.
During vertebrate evolution, the IFN ligand gene family has undergone significant expansion through both tandem gene duplication and retrotransposition events, most likely to accommodate increased regulatory diversity and functional specialization (50). Although modest, our current understanding of IFN ligand specialization is increasing. Functional specialization between major Type I, II and III ligands has been revealed by analysis of genetic mutations affecting specific receptors or downstream kinases and transcription factors in both humans and mice (41,51). For example, it is accepted that deficiencies in IFNG signaling are associated with mycobacterial disease, whereas deficiencies in Type I/III signaling confer susceptibility to viral infections (41). IFN ligand specialization is also evident in the clinical use of recombinant ligands, with IFNB1 being the most effective therapeutic agent for the treatment of multiple sclerosis, whereas IFNA2 preparations are preferred for the treatment of chronic viral infections and some malignancies (52). Despite these advances, little is known about the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint mechanisms behind these differential effects. In this context, our work provides a valuable resource for future mechanistic research. Although our multi-omics analysis is descriptive in nature and based largely on statistically significant associations that should not be interpreted as cause-effect relationships, its value is confirmed by the many associations observed for which mechanisms have already been established. For example, our unbiased analysis of the transcriptome correlations confirmed that 8 of the 12 ligands tested are indeed significantly and positively associated with a transcriptional program highly enriched for ISGs. Likewise, the association between IFNG and metabolites in the kynurenine pathway can be explained by induction of IDO1, a known ISG, during the inflammatory response elicited by SARS-CoV-2 (25)(26)(27)(28)(29). Therefore, using these confirmatory observations as reference points, we propose that the datasets described here will help the field elucidate many novel cause-effect relationships explaining IFN ligand specialization.
The specialized biosignatures of IFN ligand action described here could be due to several non-mutually exclusive mechanisms, such as action through different receptors, differences in affinity or allosteric regulation for the same receptors, as well as differences in the location and timing of ligand production.
One limitation of our study is that all measurements were performed from peripheral blood, which can only inform about a subset of the pathophysiological processes modulated by the various ligands. Our study would be highly complemented by studies of IFN ligands in various tissues (e.g. (53)). It is also possible that the specialized biosignatures observed are driven in part by SARS-CoV-2 itself. Like other members of the coronavirus family, SARS-CoV-2 has evolved diverse strategies to evade the antiviral effects of IFN signaling, and it is possible that these escape mechanisms do not affect all IFN ligands equally (54). Despite these limitations, key observations produced by our study include the differential relationship between IFN ligands and the antiviral transcriptional program in circulating immune cells, the specialized relationship between seroconversion, immune cell type abundance and IFN ligand levels, and the distinct metabolic signatures associated with the ligands. Throughout the study, the contrast between IFNA2 and IFNA6 exemplifies these points. Both IFNA2 and IFNA6 are specifically recognized by the reagents employed and significantly upregulated in the COVID19 positive cohort.
However, whereas IFNA2 is strongly associated with the IFN transcriptional program in immune cells, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. Accordingly, IFNA2 abundance associates with increased frequency of various T cell subsets involved in the early antiviral response, while IFNA6 levels correlate with signs of B cell maturation and differentiation. Whereas IFNA2 has the highest number of significant associations in the RBC metabolome of any ligand tested, IFNA6 has none. Therefore, a detailed comparative study of these two IFNA ligands is warranted, including studies in human cell preparations and animal models.
In sum, our analyses and datasets provide a rich resource to advance understanding of the IFN ligand family in humans. In order to accelerate the use of these datasets at a global scale, they are made readily available through the COVIDome Explorer Researcher Portal (covidome.org) (17), where users can rapidly recreate the cross-omics correlations described here, investigate any other cross-omics correlations of choice, and download all data for further analysis.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint Plasma proteomics by SOMAscan ® assays. 125 μL EDTA plasma was analyzed by SOMAscan ® assays using previously established protocols (55). Briefly, each of the 4000+ SOMAmer reagents binds a target peptide and is quantified on a custom Agilent hybridization chip. Normalization and calibration were performed according to SOMAscan ® Data Standardization and File Specification Technical Note (SSM-020) (55). The output of the SOMAscan ® assay is reported in relative fluorescent units (RFU). Validation of IFN detection was carried out by spiking recombinant human IFN ligands into separate aliquots of a pooled plasma reference sample (10 pg/µL). Data were processed as above and then to account for background signal in the reference sample, the median relative abundance measured by each SOMAscan ® aptamer reagent across all samples was subtracted from the corresponding values for each spike-in sample. Recombinant human IFN ligands were obtained from PBL Assay Science (Piscataway, NJ 08854 USA), with the following catalog numbers: 11002-1 (Human   Interferon Alpha Sampler Set: IFNA1, IFNA2, IFNA4, IFNA5, IFNA6, IFNA7, IFNA8, IFNA10, IFNA14, IFNA16, IFNA17, IFNA21); 11725-1 (IFNL1); 11720-1 (IFNL2); 11730-1 (IFNL3); 11500-1 (IFNG); 11420-1 (IFNB1). . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ReproSil C18 1.9 µm, 120A resin. The system was coupled to a timsTOF Pro mass spectrometer (Bruker Daltonics, Bremen, Germany) via a nano-electrospray ion source (Captive Spray, Bruker Daltonics). The mass spectrometer was operated in PASEF mode. The ramp time was set to 100 ms and 10 PASEF MS/MS scans per topN acquisition cycle were acquired. MS and MS/MS spectra were recorded from m/z 100 to 1700. The ion mobility was scanned from 0.7 to 1.50 Vs/cm 2 . Precursors for data-dependent acquisition were isolated within ± 1 Th and fragmented with an ion mobility-dependent collision energy, which was linearly increased from 20 to 59 eV in positive mode. Low-abundance precursor ions with an intensity above a threshold of 500 counts but below a target value of 20000 counts were repeatedly scheduled and otherwise dynamically excluded for 0.4 min. Raw data file conversion to peak lists in the MGF format, downstream identification, validation, filtering and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint quantification were managed using FragPipe version 13.0. MSFragger version 3.0 was used for database searches against a Human isoform-containing UniProt fasta file (version 08/11/2020) with decoys and common contaminants added. The identification settings were as follows: Trypsin, Specific, with a maximum of 2 missed cleavages, up to 2 isotope errors in precursor selection allowed for, 10 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Mass spectrometry-based metabolomics of plasma and red blood cells.
Sample extraction. Samples were thawed on ice and extracted via a modified Folch method Analysis of transcriptome data. RNA-seq data yield was ~40-80 x 10 6 raw reads and ~32-71 x 10 6 final mapped reads per sample. Reads were demultiplexed and converted to fastq format using bcl2fastq (bcl2fastq v2.20.0.422). Data quality was assessed using FASTQC (v0.11.5) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and FastQ Screen (v0.11.0, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint hyperbolic sine transformed using the cytofAsinh() function with cofactor = 5 from the cytofkit package, and 1000 cells per FCS file sampled without replacement for downstream analysis. For visualization, dimensionality reduction was performed using the t-distributed stochastic neighbor embedding (t-SNE) method from the Rtsne package (v0.15) (72), using all markers. Unsupervised clustering, using all markers, was performed using the cytofkit implementation of the PhenoGraph algorithm (22).
Transformed marker expression values for each clustered cell/event were exported and Z-scores calculated across all events for visualization on t-SNE plots. Relative frequencies for each cluster were calculated as proportions of live cells per sample for use in subsequent analyses. For traditionally gated cell subpopulations (gating strategy is described in (16)), relative frequencies were exported from CellEngine as percentages of various parental lineages for use in subsequent analyses.
Analysis of LCMS-metabolomics data. Peak intensity data was imported to R. Across the 171 metabolites, 0 values (486 missing values of 21,033 total measurements) were replaced with a random value sampled from between 0 and 0.5x the minimum non-zero intensity value for that metabolite. For downstream analysis, data was then normalized using a scaling factor derived by dividing the global median intensity value across all proteins by each sample median intensity. Median normalization was chosen as it is simple to employ, relies on few assumptions, and performs on-par with more complex normalization techniques, such as linear regression, local regression, total intensity, average intensity, and quantile normalization, in reducing intragroup variation (73), and is one of the non-reference-based normalization methods employed in the widely-used MetaboAnalyst pre-processing module (74).
Interferon Alpha/Gamma Scores. To capture interferon signaling in each sample as a single value we calculated RNA-seq-or Somascan-based 'Interferon Alpha' and 'Interferon Gamma' scores as follows: Firstly, Z-scores were calculated from the age-and sex-adjusted concentration values for each gene/protein in each sample, based on the mean and standard deviation of COVID19-negative samples. Secondly, per-sample scores were calculated as the sum of Z-scores for genes/proteins in . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint the Hallmark Interferon Alpha or Hallmark Interferon Gamma Response gene sets (18), filtered to genes/proteins with significant increases in the COVID-positive group (see next section).
Differential abundance analysis. For RNAseq, gene-level differential expression in COVID+ versus COVID-was evaluated using DESeq2 (version 1.28.1)(68) in R (version 4.0.1), with q < 0.1 (FDR < 10%) as the threshold for differentially expressed genes, and considering only genes with ≥ 0.5 countsper-million in at least two samples. Differential abundance analysis for SOMAscan ® proteomics, MSD cytokine profiling, MS proteomics, and LCMS metabolomics was performed using linear models in R packages. Extreme outlier data points (above Q3 + 3xIQR or below Q1 -3XIQR) were removed.
Beta regression analysis of MC data. To identify cell clusters or gated cell subsets for which relative frequencies are associated with plasma levels of the 12 IFN ligands in COVID19 positive samples, beta regression analysis was carried out using the betareg package (v3.1-4) (80), with each model using cell cluster/subset proportions (relative frequency) as the outcome/dependent variable and log2transformed IFN abundance values as the independent/predictor variable, with adjustment for Age and Sex, and a logit link function. Effect sizes (as fold-change per unit IFN abundance) for each IFN ligand were obtained by exponentiation of beta regression model coefficients. For comparison across IFN ligands as in volcano plots and heatmaps, beta regression model coefficients were multiplied by the standard deviation of the corresponding ligand before exponentiating to give 'standardized' fold-. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021. examples, data points were visualized as XY scatter plots, with points colored by local density using a custom function, and overlaid with beta regression fit curves and 95% confidence intervals extracted from model objects using the ggemmeans() function from the ggeffects package (v1.1.0) (81).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted August 1, 2021.

DECLARATION OF INTERESTS
JME serves in the COVID Development Advisory Board for Elly Lilly and has provided consulting services to Gilead Sciences Inc.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021.    . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint        . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 1, 2021. ; https://doi.org/10.1101/2021.07.29.21261325 doi: medRxiv preprint